HAL Id: jpa-00245591
https://hal.archives-ouvertes.fr/jpa-00245591
Submitted on 1 Jan 1987
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Impurity-defect interaction in polycrystalline silicon for photovoltaic applications. The role of hydrogen
A. Chari, P. de Mierry, A. Menikh, M. Aucouturier
To cite this version:
A. Chari, P. de Mierry, A. Menikh, M. Aucouturier. Impurity-defect interaction in polycrystalline
silicon for photovoltaic applications. The role of hydrogen. Revue de Physique Appliquée, Société
française de physique / EDP, 1987, 22 (7), pp.655-662. �10.1051/rphysap:01987002207065500�. �jpa-
00245591�
655
Impurity-defect interaction in polycrystalline silicon for photovoltaic applications. The role of hydrogen
A.
Chari,
P. deMierry,
A. Menikh and M. AucouturierLaboratoire de
Physique
desSolides, C.N.R.S., 1, place
AristideBriand,
92195 Meudon BellevuePrincipal Cedex,
France(Reçu
le 3 octobre1986, accepté
le 20janvier 1987)
Résumé. 2014 Cet article résume les études effectuées par les auteurs sur le comportement
physico-chimique
dequelques impuretés (P, C, H)
dans le silicium. Les résultats portent sur : la diffusion et laségrégation d’impuretés
dans le silicium mono etpolycristallin,
lapassivation
des défauts recombinants parl’hydrogène,
les interactions
hydrogène-dopants.
Un accentparticulier
est mis sur le comportement et la diffusion del’hydrogène.
Les résultats sont discutés en tenant compte de l’existence de mécanismescomplexes
d’interactions entre
l’hydrogène
et lesimpuretés
ou les défauts.Abstract. 2014 An overview of the studies done
by
the authors on thephysicochemical
behaviour of someimpurities (P, C, H)
in silicon isgiven.
Results concern : diffusion andsegregation
ofimpurities
in mono andpolycrystalline silicon, passivation
ofrecombining
defectsby hydrogen, hydrogen-dopant
interaction. A morefocused interest is
given
onhydrogen
diffusion and behaviour. The results arediscussed, taking
into accountthe existence of
complex
mechanisms of interaction betweenhydrogen
andimpurities
or defects.Revue
Phys. Appl.
22(1987)
655-662 JUILLET 1987Classification .
Physics
Abstracts72.80C - 61.70N - 61.70Y - 66.30J
1. Introduction.
The
development
ofpolycrystalline
silicon as amaterial for
photovoltaic applications
has raised several fundamentalquestions,
in relation with tech-nical
implications. Technically speaking,
the mainproblems arising
when this new material was pro-posed
can be summarized as follows :i)
What will be the consequence of thepolycrystal-
line nature on the
technology
of solar cell fab- rication ? In otherwords,
would thegrain-bound-
aries or other defects
(dislocation
arrays,etc.)
behave for instance as electronic of atomic diffusion short
circuits, modifying
thejunction profile
and/orthe behaviour of the cell ?
ii)
Will the defects of thepolycrystalline
materialhinder the
photovoltaic properties
of thematerial,
that is
generation
and diffusion of theminority carriers, mobility
of themajority carriers..., leading
also to a
degradation
of theefficiency
of thephotovoltaic
cells ?iii)
What would be thereliability
of the solutionsproposed
to cure this above mentioneddegradation :
specific elaboration,
modification of the process, defectpassivation by hydrogen
or other means, ... ?iv) Taking
into account all thesefactors,
whatwould be the maximum content of different defects
(point,
linear andtwo-dimensional)
and of differentimpurities
which can be allowed in a « solargrade »
material to ensure a
possible
choice of this material forphotocell fabrication ?
These technical
problems
lead to several funda-mental
questions, concerning
thephysical
andphysico-chemical
behaviour ofpolycrystalline silicon ;
one can mention for instance :- structure of the defects of the material and its influence on electronic and
photoelectric properties,
-
possible grain-boundary
fast diffusion and/orsegregation
ofdopant
andimpurities,
- nature of the interactions between
impurities
and
dopants
or defects and their consequence on electronicproperties,
- diffusion and
solubility
of thepassivating
im-purities (e.g. hydrogen)
in thismaterial,
-
passivation mechanism,
i.e. the nature of in-Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/rphysap:01987002207065500
656
teractions between
the passivant
and the recombina- tion centre.A
large
number of research work has been done andpublished, giving partial
or definitive answers to some of thesequestions [1-4].
This papersummarizes
some results obtained
by
the authors onpolycrystal-
line and
bicrystalline silicon,
which canbring
acontribution to the
knowledge
of the behaviour ofpolycrystalline
silicon in thefollowing fields :
-
Grain-boundary
diffusion andsegregation
ofthe
impurities (phosphorus,
carbon andhydrogen)
and their consequences on the recombination at
grain-boundaries,
with aspecial
interest onhydrogen passivation
ofbicrystals.
- Introduction and diffusion of
hydrogen
intosilicon.
- Interactions between
hydrogen
andimpurities, especially dopants,
and consequences on electronicproperties.
2.
Grain-boundary
diffusion andségrégation
of im-purities.
2.1 PHOSPHORUS. - The
preferential
diffusion ofphosphorus
in silicongrain-boundaries
mayhave,
ifit
exists, important
consequences on the diffusedjunction profile
of thephotovoltaic
cells[5]. System-
atic measurements were conducted on
p-type (B doped,
p = 1Ocm) polycrystalline
silicon elaboratedby
the R.A.D. process[6]
orby ingot casting (Wacker Heliotronic).
A radioactivation method was used :
phosphorus
is diffused into the material
by
the conventional borosilicateglass
process, at varioustemperatures,
itis activated
by
neutron bombardment and the pro- files areanalysed by grinding
andresidual activity
measurements
[7] ;
the results are summarized infigures
1 and 2. Bulk diffusion coefficients of P in Siare
measured,
but nograin-boundary preferential
diffusion of P could be observed in this kind of
material,
neither from thepenetration
curves, nor from the observation ofautoradiography.
The absence of
grain-boundary
diffusion observedhere is in
disagreement
with results obtainedby
other laboratories
[5, 8]
ondifferent
materials. The reason is that R.A.D. andingot
castpolycrystalline
silicon contain a
large majority
ofgrain-boundaries
in coincidence or near from coincidence.
The measurements of
phosphorus intergranular diffusivity
mentioned[8]
were obtained inrecrystal-
lized silicon with
mostly
randomgrain-boundaries.
2.2 CARBON. - Carbon is often a
majority impurity
of solar
grade polycrystalline
silicon. Astudy
ofcarbon diffusion and
segregation
has been conductedon R.A.D. and Wacker
polycrystalline
silicon. A radioactive tracer method is used : thespecimens
arefirst coated with a
layer
of radioactive(i5C labelled)
Fig.
1. - Variation of the 32p residualactivity (03B2)
as afunction of the
depth
inpolycrystalline
silicon. Diffusion anneal : T =900 °C ;
t = 5h ;
Nograin-boundary
diffu-sion tail is observed, and the full curve follows the
ILAn - _X = K exp (- X2n/4 Dt)
lawexpected
for a purebulk diffusion.
Fig.
2. - Arrheniusplot Log (D)
=/(l/7r)
for bulkdiffusion of
phosphorus
intopolycrystalline
silicon.7
uC
layer by annealing
at moderatetemperature
)00°C)
in a sealed silica tubecontaining
thea14 C03 compound.
Various heat treatments are ten done and the14C profiles
areanalysed by inding
and residualactivity
measurements[7].
For[e
segregation studies,
thespecimens
are firstmogeneized by prolongated
treatment athigh
mperature
in the presence ofBa 14 C03 (1200 °C, N x ) days)
and then annealed at varioustemperatures S
provoke
theexpected segregation.
Thé segrega-on is detected
by autoradiography (conventional d
id
high resolution)
of thespecimens [7].
The results on diffusion are summarized in
%ures 3 and 4. Bulk diffusion
coefficients
of carbon to silicon are obtained at differenttemperatures
it as for
phosphorus
nograin-boundary preferential
ffusion could be detected in this kind of
material,
;ither
by analysis
of thepenetration
curves, norby itoradiography.
Furthermore a
strong segregation tendency
ofrbon to
grain-boundaries
has beenobserved.
NotFig.
l the
grain-boundaries
are affectedby
this segrega- diffan
phenomenon (Fig. 5).
In order to
try
to correlate thesegregation ability
.th the recombination behaviour of the
grain- nndaries,
crossedexperiments
are conducted asllows
[7, 9].
Acartography
ofminority
carrierscombination centres is done
by
E.B.I.C.(Electron
;am induced
current)
ongiven specimens
and théme
specimens
are thenhomogeneized
with14C
and500 micrc
radio of th
250 a
of th
25 50 75 100
anneX (CM) x 10-4
ThéThe
3.
- Variation of the 14C residualactivity (13)
as a thatiction of the
depth
inpolycrystalline
silicon ; diffusion wher neal : T = 1 353°C ; t
= 2 h 30 min. The(b)
part of the Inrve is
perturbed by
carbideprecipitation phenomena.
samele
(a)
part follows theA ~An ~Xn = (- X2n/4 Dt
law ean 1give pected
for a pure bulk diffusion. small10°-
-8,1. -9
10
1io- 10
il 19 .
6 6,5
7 T-1 x 104 (°k-1)
4. - Arrhenius
plot Log (D)
=f(1/T)
for bulksion of carbon into
polycrystalline
silicon.5. -
Autoradiography
of14C coupled
with EBICe on a
specimen
of R.A.D. Black dots(optical )graph)
or white dot(S.E.M. micrograph)
of thegraphy
indicate the presence of14C.
The black lines.e E.B.I.C.
image
indicaterecombining
boundaries.,aled.to obtain the
grain-boundary segregation.
comparison
between theautoradiography
ofand the E.B.I.C.
images
showsclearly (Fig. 5)
the
recombining
boundaries are also boundaries^e a
segregation
of carbon is observed.an electron
microscopy investigation
of thematerial, Sharko [10]
has shown that carbon beprecipitated
as C-Simicroparticles
in somei
grain-boundaries (mostly
boundaries with a 1 deviation tocoincidence).
Both our andSharko’s results agree to prove that carbon seg]
tion or
precipitation
isresponsible
for the min carrier recombination behaviour of these g boundaries.2.3 HYDROGEN IN POLYCRYSTALS. - The
hydr
introduction and bulk diffusion
aspect
will be tailed in afollowing
section 3. Theproblem
o:interaction between
hydrogen
andgrain-bound
is of
great importance
to understand the mechar of the recombination centrepassivation by
element.
A radiotracer
technique
has been also used incase : tritium
(3H )
is introduced intop-type poly
talline silicon
(R.A.D.
orWacker) by
catlpolarization
in acidic salt bathelectrolyte
at Iflabelled with tritiated water
[11]. By photocu:
measurement it is checked that a certain
passiv
of the
recombinating
centres is thus obtainedHigh
resolutionautoradiographies (0.2
03BCm rution)
areexposed
on thespecimens
either inbulk
shape
or as thin foilsready
for transmit electronmicroscopy [13].
The observation ofautoradiography,
either in thescanning
ele(microscope
or in the transmission electron mscope
(Fig. 6),
shows astrong interaction
ofhi
gen with some
given grain-boundaries,
and 1precisely,
withlinear
defects(«
extrinsic çcations »)
of theseboundaries.
So thesegreg:
behaviour of
hydrogen
isthen
verysimilar
to th otherimpurities.
2.4 HYDROGEN IN BICRYSTALS. - In order to t understand more
completely
the role of g boundariesimpurities interaction,
in the recoml tion andpassivation phenomena,
asystematic
was done on
bicrystalline specimens.
The mateia 03A325 (710) bicrystal
elaboratedby
Czochiprocess. It is known
[13]
that thisgrain-bour
their From
the results(Fig. 7)
thefollowing
obse)S10n tions can be drawn : the
Fig.
7. - L.B.I.C.scanning (electrolytic diode)
gsilicon
bicrystal (03A3
= 25,(710))
after different heat t]ment, before
(2013)
and after(- - - -) plasma hydro
.m on ation. Incident
light :
GaAs laser(A
= 0.954 03BCm ; 03B1- in a 50 03BCm ; beam diameter 3003BCm).
E. M.
itium.
indi-
The
preceeding
results[13]
on the role ofthermal treatment are confirmed. The boundai
slightly recombining (recombination
rate s =103
cm.s-1)
after 450 °C andstrongly
recombi]try
to(s
= 2.5 x104
cm.s-1)
after 750 °C and 900 °C;rain- nealing.
Thehydrogen
treatment leads to a comf bina-passivation
of the 450 °C annealedboundary
butstudy
no influence on the behaviour of the 750 °C rial is 900 °C annealedspecimens.
Thepenetration
ralski
hydrogen
has been checkedby
S.I.M.S.(seconc
idary
ion massspectroscopy) profiling
after deuter659
plasma annealing.
This very recent result of astudy
still under continuation proves that :
i)
theimpurity segregation responsible
for the creation of recombin-ing
centres is notunique,
as the 450 °Cannealing has
not the same effect as the 750 °C or 900 °C anneal-
ing ; ii) hydrogen
does notpassivate
all the recombi- nation centres and interactspreferentially
with thecentres created at 450 °C. An
autoradiography study
after tritium introduction is to be done on those
bicrystals.
3. Introduction and diffusion of
hydrogen
intosilicon.
Hydrogen
can be introduced into siliconby
severalmeans : bombardment in a Kaufman source
[14, 15], plasma annealing [16], electrolyte charging [12].
Plasma
annealing
andelectrolyte charging
have beeninvestigated
in thepresent study,
in order toquantify
the diffusion mechanisms of
hydrogen.
For thisstudy, monocrystalline silicon
is used as a firststep.
3.1 HIGH TEMPERATURE INTRODUCTION BY PLAS- MA ANNEALING. - The diffusion
profiles
ofhydro-
gen into silicon at
high temperature
are then studiedby
S.I.M.S. of deuterium(hydrogen
isreplaced by
deuterium in the
plasma atmosphere (1 mbar)).
Twodifferent
plasma
sources were used : a R.F.plasma (13.56 MHz,
20W, specimen
inside theglow
dis-charge)
and a microwaveplasma (2
450MHz,
60W, specimen
outside theglow discharge).
Some results are
given
infigures
8 and 9 at varioustemperatures
and for various materials :p-type (6
x1016
at. B .cm-3), n-type (1018
at. P.cm-3)
andundoped.
From this non-achievedinvestigation,
some
provisory
conclusions may be drawn :a)
thesolubility
ofhydrogen
in silicon is in all case, « reverse », thatis, decreasing
when the tem-perature
is raised.b)
theapparent solubility
forgiven
conditions(i.e.
for agiven plasma source),
isstrongly depen-
dent on the
doping type
anddoping
level. It is oneorder of
magnitude larger
inp-type
silicon than in n-type
orundoped
silicon.c)
the diffusionprofiles
cannot beanalysed by
asimple
diffusionmechanism ;
each of them is the result of the additionof,
atleast,
two diffusionprofiles.
The observation of suchcomplex
diffusionprofiles
has beenalready
mentioned in studies on p-type
silicon[17, 18],
and theexplanation usually given
is the interactionphenomena
betweenhydro-
gen and boron
(see following
Sect.4).
d)
the maximum values of the diffusion coeffi- cients which can be deduced from thedeepest part
of theprofiles
are of the order of10-12 cm2. s-1
at150 °C and
10- Il cm2. s-1
at320 °C,
inagreement
with the literature[19].
It must be mentioned thatFig.
8. - Deuterium S.I.M.S.profils
in p-type(6 1016 at.B.cm-3)
in silicon after various deuteriumplasma annealings. a)
R.F.plasma ; (1)
T =150 °C,
t = 3
h ; (2)
T = 260°C, t
= 3 h.b) microwave plasma ; (1)
T =150 °C ; t = 3 h ; (2) T = 320 °C ; t = 3 h.
these values concern
hydrogen
detectable for thegiven
conditions of S.I.M.S.analysis (detection
limit1015
to 5 x1015 at . cm- 3 of deuterium).
3.2 ELECTROCHEMICAL PERMEATION OF HYDRO- GEN. - The electrochemical
permeation technique
has been used to detect very small fluxes of
hydrogen diffusing through
a silicon membrane at room tem-perature [20].
In this kind ofexperiments,
a thin(~ 100 03BCm)
membrane ofmonocrystalline
silicon(p-
type
orn-type
with a n+/p junction
on onesurface)
is covered on both sides
by
apalladium layer.
The« entry
face » isexposed
to an acidic solution(H2S04,
1N)
andcathodically polarized
toproduce
atomic
hydrogen (constant
current 10mA/cm2).
The« down stream face » is
exposed
also to an elec-trochemical cell
anodically polarized (V
=100
mV/ECS)
inH2S04,
1 N. The anodic currenti A
measured in the « down stream » cell is a measureof the flux of
hydrogen,
reoxidized in thiscell,
whichhave crossed the membrane. The
palladium coating
ensures the current
transport.
Thepermeation
curvei A
=f(t) gives
the variation of thehydrogen
flux as660
Fig.
9. - Deuterium S.I.M.S.profiles
in silicon after deuterium microwaveplasma annealing ; a)
n type(1018 at.P.cm-3) ; (1) T = 150°C ;
t=3h;(2)
T=320 °C ; t
= 3 h ;b) undoped
silicon ;(1)
T= 150 °C ; t = 3 h ; (2) T = 320 ° C ; t = 3 h.
a function of the time
(Fig. 10).
The mathematical deconvolution of this curve[21]
leads to theapparent
.diffusion coefficient of
hydrogen
at roomtempera-
ture. The
thermodynamical
conditions are such that thethermodynamic fugacity
ofhydrogen
at the«
entry
face » is veryhigh (larger
than10 000
atm.) [21].
As often in this kind of exper-iments,
the mathematicalanalysis
of thepermeation
curve shows that the Fick’s law for diffusion is not
satisfied. Such effect is
usually
a consequence ofcomplex
diffusionphenomena,
more or less hinderedby trapping
ofhydrogen by
the surface or defects in the material[21].
In thepresent
case, it waspossible
to show that the
permeation
curve offigure
10 canbe described
by
two sets ofapparent
diffusion coefficients. One set of values(« strongly-trapped » hydrogen)
are of the order of10-10 cm2. s-1,
theother
set(«
diffusible »hydrogen)
is of the order of10-9 cm2.s-1 [22].
Both values are orders of mag- nitudelarger
thanexpected
from theextrapolation
to room
temperature
of the coefficients obtained athigh temperature [19].
This result meansagain
thatFig.
10. -Hydrogen permeation
flux normalised tosteady
statepermeation
flux,through
100 )JLm thick mem-brane of p-type silicon as a function of time. Electrochemi- cal introduction and detection of
hydrogen
at room temperature.hydrogen
diffusion is a verycomplex phenomenon.
Under cathodic
charging, i. e. ,
under veryhigh
",fugacities
of atomichydrogen,
small amounts of thiselement can diffuse at rates much
higher
than thespecies
detectedby
S.I.M.S.analysis
of theplasma
treated
specimens.
4.
Hydrogen dopant
interaction.Hydrogen
introduced inp-type
silicon has anotherimportant
effect : forlarge quantities
ofhydrogen,
one observes a neutralization of the
dopant, leading
to a
spectacular
increase of the near-surface resistivi-ty
and astrong
decrease of themajority
carrierconcentration
[16-18].
This effect has been observed and measured inp-type
silicon submitted to thehydrogen plasma. Profiling
of themajority
carrierconcentration under the surface is obtained
by
measurement of the
voltage dependence
of thecapacity
of a diode between mercury and thespecimens.
To extend theprofile
atlarge
concen-tration,
successiveetchings
of the surface are done.The results
(Fig. 11)
confirm théstrong
neutraliza- tion effect ofhydrogen
on thedopant,
in theregions
where
large quantities
ofhydrogen
have diffused.The neutralization
depth
is ingood agreement
with
the
concentrationprofiles
of deuterium measuredby
S.I.M.S.The
physical
mechanism of thedopant
neutraliza- tion is still under discussion[23]
but it is clearthat,
from the
physicochemical point
ofview,
suchstrong interaction
willprovoke important
modifications of the diffusionmechanisms.
We are heretypically
inthe case of
strong trapping phenomena.
5. Général discussion and conclusion.
The
problem
ofimpurity-impurity
andimpurity-
defects interactions in silicon and their consequences
on electronic and
photoelectrical properties
ofpoly-
661
Fig.
11. -Acceptor profiles
from the surface in p-type silicon(6
x1016 at.B.cm-3)
afterhydrogenation plasma annealing (320 °C,
6 h ;150 °C,
3h).
crystalline
silicon is far to becompletely solved.
Theresults
given
here indicate someinteresting
trends :a)
No noticeablegrain-boundary
diffusion of thedopant
exists in thestrongly
texturedlarge grain
materials
developed
forphotovoltaic applications.
b)
Theproblem of impurity segregation,
alsodependent
on thegrain-boundary
structure, is ofmajor importance
to understand the recombination behaviour of the boundariesand, probably,
otherdefects
(dislocations) [23].
c)
Thehydrogen passivation
of the recombinationcentres cannot be
only
described in terms of structur- al defect interaction withhydrogen,
as often done(e.g. dangling
bond saturationby hydrogen) ;
themicrochemical
aspect
has to be taken into account,as recombination centres of chemical
origin
are alsopassivated by hydrogen.
d)
Thisproblem of hydrogen passivation
is be-coming
even morecomplex
if one takes into accountthe
hydrogen dopant
interaction. The fact thathydrogen
is able to neutralize with a veryhigh efficiency
theacceptors
must have an influence onthe passivation
mechanisms.e) Hydrogen
diffusion into silicon is not asimple
mechanism. At least two or three
species
ofhydro-
gen exist in the material with
respectively decreasing
solubilities and
increasing
diffusivities.Outgasing (exodiffusion) experiments
as a function oftempera-
ture, andhigh temperature permeation experiments
are necessary to
separate correctly
the behaviour of thesespecies.
At thepresent state,
thehydrogen- dopant
interaction seems to be thepredominant
factor
influencing
the diffusionphenomena.
Acknowledgments.
This
study
waspartly
financedby
COMES(Commis-
sariat à
l’Energie Solaire),
AFME(Agence Française
pour la Maîtrise de
l’Energie)
and PIRSEM(Projet interdisciplinaire
de Recherche du CNRS sur l’Ener-gie
et les MatièresPremières).
The authors are
grateful
to J. Chevallier(CNRS, Meudon)
and N. Proust(Thomson, Orsay)
for theaccess to
hydrogen
and deuteriumplasma equip-
ments.
References
[1]
Coll. Int. Semiconducteurspolycristallins, Perpignan,
J.
Physique Colloq.
43(1982)
C1.[2] Proceeding
of the Int. School of Mat. Sci. and Tech.The Ettore
Majorana
CentreErice, Italy, July
1-15
(1984) (Springer)
1985.[3]
M.R.S. Fall,Meeting,
Boston 1985, M.R.S.(1986).
[4]
Second M.R.S.Europe
Conferencepoly-micro-crys-
talline and
amorphous
semiconductors(June 1984), Strasbourg (France)
Editedby
P. Pinardand S. Kalbitzer
(Editions
dePhysique)
1984.[5] BAUMGART,
H., LEAMY, H. J.,CELLER, G. K.,
TRIMBLE, L. E., J.Physique Colloq.
43(1982)
C1-363.
[6]
BELOUET, C., HERVO, C.,MAUTREF,
M., PAGES, C., HERVO, J., 16th IEEE, Photonet.Spec.
Conf.
(1982)
SanDiego (USA)
p. 80.[7] CHARI,
A., Thèse de 3eCycle, Orsay (1980).
[8] LIOTARD,
J. L., BIBERIAN, R. andCABANE,
J., J.Physique Colloq.
43(1982)
213.[9]
AUCOUTURIER, M.,CHARI,
A.,PIRDES,
14thIEEE
Spec.
ConferenceJanuary (1980)
St.Louis
(USA) p. 1192.
[10]
SHARKO, R.,GERVAIS,
A. andTEXIER-HERVO,
C., J.Physique Colloq.
43(1982)
129.[11]
RALLON, O., AUCOUTURIER, M., HERVO, C., MAUTREF, M., BELOUET, C., Solar Cells 9(1983)
149.[12] AUCOUTURIER,
M., RALLON,O.,
MAUTREF, M. and BELOUET, C., J.Physique Colloq.
43(1982)
C1-117.
662
[13]
BATTISTELLA, F., ThèseDocteur-Ingénieur
Toulouse
(1985).
[14]
MAUTREF, M., LACROIX,C.,
BELOUET,C.,
FAGES, C., BIOTTEAU, B., ARNOULT, F., RevuePhys.
Appl.
19(1984)
333.[15] SEAGER,
C. H., SHARP, D. J., PANITZ, J. K. L., HANOKA, J. I., J.Physique Colloq.
43(1982)
C1-103.
[16]
PANKOVE, J. L., VANCE, R. O. and BERKEYHEISER, J. E.,Appl. Phys.
Lett. 45(1984)
1100.[17] PANKOVE,
J. L., MAGGEE, C. W. and VANCE, R.O., Appl. Phys.
Lett. 47(1985)
748.[18]
JOHNSON, N. M.,Phys.
Rev. B 31(8) (1985)
5525.[19]
PEARTON, S. J., J. Electron. Mater. 14a(1985)
737.[20]
CHARI, A., MIERRY, P. de, AUCOUTURIER, M. andGOROCHOV,
O., 7thEurop.
Conf. Surf.Sci.,
Aix en Provence
(France) (1984).
[21]
BRASS, A. M., These Docteur ès Sciences,Orsay (1983).
[22]
MIERRY, P. de and AUCOUTURIER, M., to bepub-
lished.
[23] CAPIZZI,
M.,MITTIGA,
A., 2nd Int. Conf. on shallowImpurity
centers, Trieste,July (1986).
[24]
ZEHAF, M., Thèse de Doct. èsSciences,
Marseille(1986).